RecBole

RecBole copied to clipboard

DeosGCF implementation

4

Paper here: https://arxiv.org/abs/2011.02100 TF implementation here: https://github.com/JimLiu96/DeosciRec The algorithm is pretty similar to LightGCN so there wasn't too much to do. The paper claims an improvement over LightGCN, which I'm...

deklanw

[💡SUG] Automatic dataset and dataloader save path

1

For my understanding there are two different level of dataset preprocessing, Level 1 **filtering and preprocessing operation**: you can save a dataset after this step calling `dataset.save()` Level 2 **dataset...

damicoedoardo

[💡SUG] Case Study Extension -- Report rank position on each user

2

Currently, the performance is reported by taking the average across the eval dataset, but I find it is challenging to do a case-by-case comparison between my experimental group and the...

rowedenny

[💡SUG] Metrics for each session

1

**Is your feature request related to a problem? Please describe.** To make the significance testing, we need to outcome the metrics for each session separately and then do the significance...

rowedenny

[🐛BUG] Load pop model and predict, throw follow exception, self.max_cnt is none.

1

topk_score, topk_iid_list = full_sort_topk(uid_series, pop_model, test_data, 20) File "/data/home/anaconda3/envs/pytorch/lib/python3.7/site-packages/recbole-0.2.1-py3.7.egg/recbole/utils/case_study.py", line 87, in full_sort_topk scores = full_sort_scores(uid_series, model, test_data) File "/data/home/anaconda3/envs/pytorch/lib/python3.7/site-packages/torch/autograd/grad_mode.py", line 26, in decorate_context return func(*args, **kwargs) File "/data/home/anaconda3/envs/pytorch/lib/python3.7/site-packages/recbole-0.2.1-py3.7.egg/recbole/utils/case_study.py", line...

dxjjhm

[💡SUG] It seems that RecBole can't run on multi-gpu in parallel

1

I find that RecBole can't run on multi-gpu in parallel, will parallel training be implemented in the future?

KinvenW

[💡SUG] 对数据集的优化

3

在一个15G的 .inter 用户数据做测试，内存使用达到了130+G. 在数据的处理上，dataset.py 是否考虑做进一步的优化？

dxjjhm

[💡SUG] Performance at different levels of activity: "cold-start" vs "hot"

4

It would be convenient to disentangle performance for "cold-start" and "hot" situations for both users and items. Interesting example for the user case from "Noise Contrastive Estimation for One-Class Collaborative...

deklanw

[💡SUG] Distributed train and evaluation

1

I wonder if RecBole will support distributed pytorch architecture such as horovod?

suhmily

[💡SUG] 实现并开源这么一个伟大的项目真是太给力了！但是没有review-based-recommendation的常用模型吗？

3

PanTings